Voice · google/gemini-3.1-flash-tts
Gemini 3.1 Flash TTS — speech with intonation set in words
Gemini 3.1 Flash TTS is Google’s text-to-speech model, chosen for its expressiveness: you set the delivery with a plain phrase and add a sigh or a laugh with a tag right in the text. The model has 30 voices and more than 70 languages; limko offers a selection of 12 voices and 17 languages, including Russian, Ukrainian and English. Voice-over costs 24 ₽ per 2,000 characters, and it is the one to take when the voice needs to act, not just read.
from 24 ₽ per voice-over up to 2,000 chars
How Gemini TTS stands out from its neighbors
- Delivery by instruction: limko’s eight ready-made delivery styles — from calm to ad voice — are turned into a phrase the model understands.
- Tags in the text: [sigh] and [laughing] add a sigh and a laugh exactly where they stand in the script.
- 12 voices with character: six female and six male — from confident and warm to clear and gravelly.
- 17 languages on the list — from Russian and Polish to Arabic, Hindi and Hebrew.
- A short text is usually ready in 10–20 seconds; the recording waits in your history, so you can close the page.
Which voice to pick for the task
In limko you get 12 of the model’s 30 voices. The pairs below are a starting point: compare two or three voices on a short fragment, since a line shorter than 300 characters is billed as 300.
| Task | Female voice | Male voice |
|---|---|---|
| Presentation, report | confident | firm |
| Ad, announcement | bright | energetic |
| News, review | breezy | narrator-like |
| Lesson, instructions | soft | clear |
| Story, audio story | warm | gravelly |
| Video for a young audience | youthful | upbeat |
Delivery in words, emotions with tags
Gemini’s delivery is set not through settings but through an instruction in natural language. You do not have to write it yourself: pick a delivery style — “Neutral”, “Calm”, “Cheerful”, “Excited”, “Whisper”, “Sad”, “Narrator” or “Ad voice” — and limko passes a ready-made wording to the model.
Pinpoint emotions go in as a tag in the text itself. The model understands tags like [sigh] and [laughing]: in the line “Well, that’s it [sigh] — time to say goodbye”, the sigh lands where the tag is, and the rest of the paragraph sounds even.
17 languages and a limit in bytes
The limko list includes Russian, English, Ukrainian, German, French, Spanish, Italian, Portuguese, Polish, Turkish, Azerbaijani, Chinese, Japanese, Korean, Arabic, Hindi and Hebrew. Voice and language are chosen separately, so versions of a video in different languages can be voiced with the same timbre.
The model counts its text limit in bytes, not characters. In Cyrillic about 1,900 characters fit into one run, in Latin script roughly twice as many. A long script is split into parts by scene or paragraph, and the finished recordings are merged and trimmed for free right in limko.
Gemini or another text-to-speech model
Gemini’s expressiveness costs 24 ₽ per 2,000 characters. That price pays off for ads, storytelling, emotional lines and multilingual versions.
A long article or chapter is cheaper to hand to Grok TTS: 2.2 ₽ per 2,000 characters and up to 14,500 characters per run. Studio sound for a course or audiobook comes from MiniMax Speech HD, and voicing a text in your own voice is possible only in Qwen3-TTS.
Frequently asked about Gemini TTS
How is Gemini TTS different from Grok TTS?
Gemini is more expressive: delivery is set by an instruction, emotions by tags, and it has 12 voices in limko versus five. Grok TTS is cheaper and takes up to 14,500 characters per run — it is the one for long texts.
How much does voice-over in Gemini TTS cost?
24 ₽ per 2,000 characters; a longer text costs proportionally more, and the shortest line is counted as 300 characters. The exact total is shown on the button before you start, and credits are charged only for a successful voice-over — if something fails, they return to your balance.
Does Gemini TTS speak Russian?
Yes, Russian is on the list along with Ukrainian, English and 14 more languages. The “Prepare for voice-over” button adds stress marks and the letter ё to the text in advance — for a small separate charge.
How much text fits into Gemini TTS in one run?
About 1,900 characters in Cyrillic: the model counts bytes, and a Russian letter takes twice as many of them as a Latin one, so roughly twice as much English text fits. A long text is voiced in parts, and the recordings are merged in limko for free.
Can I voice a text in my own voice?
Not in Gemini TTS — it has 12 preset voices. In limko, voice cloning is available only in Qwen3-TTS.